利用ARIMA模型对医院感染现患率进行预测

Prediction of prevalence rate of nosocomial infection by using ARIMA model

目的利用差分整合移动平均自回归(Autoregressive Integrated Moving Average,ARIMA)模型对医院感染现患率进行预测,为防控全院的医院感染提供决策依据。方法回顾性调查收集某院2012年1月-2017年12月(共72个月)的医院感染现患率,采用ARIMA时间序列模型建模,并评价模型拟合和预测效果。结果经过比较备选模型,最终确认ARIMA(1,0,0)模型拟合效果相对较优,2018年前6个月现患率实际值均在预测值的95%可信区间内。结论该模型能准确预测医院感染的现患率,有助于医院感染管理相关预防控制措施的制定与实施。

OBJECTIVE The ARIMA model was used to predict the prevalence rate of nosocomial infection and to provide decision making basis for the prevention and control of nosocomial infection in the whole hospital. METHODS A retrospective survey was conducted to collect the prevalence rate of nosocomial infection in a hospital from Jan. 2012 to Dec. 2017(a total of 72 months).A model was constructed by ARIMA time series model, and the fitting and prediction effects of the model were evaluated. RESULTS After comparing the alternative models,it was confirmed that the fitting effect of ARIMA (1,0,0)model was relatively good. The actual prevalence rates in the first six months of 2018 were within 95% confidence interval of the predicted values. CONCLUSION The model can accurately predict the prevalence rate of nosocomial infection, and is helpful to formulate and implement the preventive and control measures related to nosocomial infection management.